Employment correlates positively with health, but is employment cause or consequence? The social causation hypothesis says that employment improves the health of men and women. The selection hypothesis says that healthy people get and keep jobs more than unhealthy people do. We test both hypotheses using longitudinal data from a national probability sample (N = 2,436 interviewed in both years). In the equations representing social causation, full-time employment predicts slower declines in perceived health and in physical functioning for both men and women. Full-time employment has the same effect for both sexes. Among women, it also has the same effect for White and non-White, and for married and nonmarried. In the equations representing social selection, physical functioning increases the odds of getting or keeping a full-time job for both sexes. Perceived health increases the odds for women but not for men. In regard to homemaking among women, homemaking predicts significantly greater declines in health, but health has no effect on the odds of becoming or staying a homemaker.